Opposing Effects of Reduced Kidney Mass on Liver and Skeletal Muscle Insulin Sensitivity in Obese Mice

Metabolic disorders such as obesity, diabetes, and dyslipidemia may lead to progressive renal damage. Conversely, there is increasing evidence that reduced kidney function may deteriorate glucose metabolism and insulin sensitivity in humans (1,2). In addition, live kidney donation may increase the risk of developing insulin resistance and metabolic syndrome (3,4). Such clinical observation is supported by recent experimental studies in rats showing the development of glucose intolerance, dyslipidemia, and ectopic fat accumulation in parallel with the development of uremia in uninephrectomized animals (5,6). Mechanistically, overnutrition may activate the renin-angiotensin system (RAS) and thus may contribute to the pathogenesis of the metabolic syndrome in the presence of reduced renal function. In particular, activation of the RAS/angiotensin receptors impairs insulin signaling in adipose tissue, skeletal muscle, and liver (7), and its prevention by angiotensin receptor blockade (pharmaceutically or genetically) improves glucose homeostasis (8,9). In particular, the angiotensin II type 1 receptor blocker telmisartan was previously reported to improve obesity-associated/high-fat diet (HFD)–associated adipose tissue inflammation (10–12). Moreover, uninephrectomized rats developed fat redistribution that could be prevented by treatment with an ACE inhibitor (6).

Skeletal muscle is the major site of insulin-induced glucose disposal in the postprandial state. Besides directly stimulating glucose uptake into muscle fibers, insulin increases microvascular blood flow to the muscle, accounting for approximately half of the insulin-mediated glucose uptake (13). Accordingly, the number of capillaries perfusing the muscle is positively related to peripheral insulin action (14). Moreover, reduced blood flow to the muscle is correlated with insulin resistance (15–18), and insulin-resistant humans and rodents exhibit capillary rarefaction (19–21). However, it still remains unclear whether the reduction in capillaries is a cause or consequence of muscle insulin resistance (22).

In the current study, we sought to determine the impact of reduced kidney mass as established by uninephrectomy (UniNx) on glucose metabolism in HFD-fed mice. Unexpectedly, we found that UniNx resulted in decreased adipose tissue inflammation and improved hepatic insulin sensitivity and steatosis. In contrast, UniNx reduced skeletal muscle capillary density and deteriorated muscle insulin action in vivo.

Male C57BL/6J (C57BL/6JOlaHsd) mice were purchased from Harlan (AD Horst, the Netherlands). All mice were housed in a specific pathogen-free environment on a 12-h light-dark cycle (light on from 7:00 p.m. to 7:00 a.m.) and fed ad libitum with regular chow diet (Provimi Kliba, Kaiseraugst, Switzerland) or HFD (58 kcal% fat w/sucrose Surwit Diet, D12331, Research Diets). All protocols conformed to the Swiss animal protection laws and were approved by the Cantonal Veterinary Office in Zurich, Switzerland.

Male C57BL/6J mice underwent UniNx or sham operation at 7 weeks of age. Mice were anesthetized with isoflurane (Abbott, Baar, Switzerland). They were placed on a warming pad and kept under an infrared heating lamp to stabilize body temperature during the whole surgical procedure. Left nephrectomy was performed through a 1.0-cm incision on the left dorsolateral paralumbar region as follows. After skin incision, abdominal muscles were incised to expose the retroperitoneal region. The left kidney was secured with a clamp (Francis chalazion forceps, D-8425), and fat attached to the kidney was removed. Special care was taken to prevent damage to the adrenal gland. Renal blood vessels and ureter were ligated with sterile silk surgical sutures. Subsequently, the left kidney was excised distal to ligatures. Abdominal muscles were sewn with absorbable thread, and the opposite ends of incised skin were clipped together using sterile disposable skin staplers. Sham-operated control mice underwent an identical surgical procedure except for kidney removal. Subcutaneous injection of buprenorphine (Essex, Luzern, Switzerland) every 6 h for 2 days was used for analgesia.

Mice were fasted overnight for intraperitoneal glucose tolerance tests and for 3 h for intraperitoneal insulin tolerance tests. Either glucose (2 g/kg body weight) or human recombinant insulin (1.0 units/kg body weight) was injected intraperitoneally (23).

Glucose clamp studies were performed as previously described (24). Clamps were performed in freely moving mice. Glucose infusion rate was calculated once glucose infusion reached a more or less constant rate with blood glucose levels at 5 mmol/L (80–90 min after the start of insulin infusion). Thereafter, blood glucose was kept constant at 5 mmol/L for 20 min, and glucose infusion rate was calculated. The glucose disposal rate was calculated by dividing the rate of [3-³H]glucose infusion by the plasma [3-³H]glucose–specific activity (25). Endogenous glucose production during the clamp was calculated by subtracting the glucose infusion rate from the glucose disposal rate (25,26). Insulin-stimulated glucose disposal rate (IS-GDR) was calculated by subtracting basal endogenous glucose production (equal to basal glucose disposal rate) from glucose disposal rate during the clamp (27). In order to assess tissue-specific glucose uptake, a bolus (10 μCi) of 2-[1-¹⁴C]deoxyglucose was administered via catheter at the end of the steady state period. Blood was sampled 2, 15, 25, and 35 min after bolus delivery. Area under the curve of disappearing plasma 2-[1-¹⁴C]deoxyglucose was used together with tissue concentration of phosphorylated 2-[1-¹⁴C]deoxyglucose to calculate glucose uptake, as was previously described (28).

Locomotion, food intake, O₂ consumption, and CO₂ production were determined for single housed mice during a 24-h period in a metabolic and behavioral monitoring system (PhenoMaster, TSE Systems, Bad Homburg, Germany) as previously described (29).

Plasma insulin and free fatty acid (FFA) levels were determined as previously described (23). Plasma angiotensin I levels were determined by an ELISA kit (Cusabio Biotech Co. Ltd., Wuhan, China). Serum creatinine levels were measured using a DetectX Low Sample Volume Serum Creatinine kit (K0021-H1D, Arbor Assays, Ann Arbor, MI), serum uric acid levels by a QuantiChrom Uric Acid Assay kit (DIUA-250, BioAssay Systems, Hayward, CA), and plasma bile acid levels using a Mouse Total Bile Acids Assay kit (80470, Crystal Chem Inc., Downers Grove, IL).

Mice were fasted for 4 h prior to the analysis. Extensor digitorum longus (EDL) and soleus muscles were removed from anesthetized mice (Avertin, 99% 2,2,2-tribromo ethanol, and tertiary amyl alcohol at 15–17 μL/g body weight i.p.) and incubated for 30 min at 30°C in vials containing preoxygenated (95% O₂-5% CO₂) Krebs-Henseleit buffer (KHB) containing 5 mmol/L HEPES (prebuffer) and supplemented with 15 mmol/L mannitol and 5 mmol/L glucose. Muscles were transferred to new vials containing fresh pregassed KHB as described above with or without 120 nmol/L of insulin (Actrapid, Novo Nordisk, Mainz, Germany). Afterward, muscles were transferred to new vials containing preoxygenized KHB supplemented with 20 mmol/L mannitol and incubated for 10 min. Muscles were then transferred to new vials containing preoxygenized KHB supplemented with 1 mmol/L [³H]2-deoxy-glucose (2.5 μCi/mL) and 19 mmol/L [¹⁴C]mannitol, to account for extracellular space, and incubated for 20 min. After the last incubation, muscles were washed in ice-cold KHB, dried externally on filter paper, and quickly frozen with aluminum tongs precooled in liquid nitrogen and stored at −80°C. Glucose transport rates were determined by scintillation counting of cleared protein lysates as described (30).

For assessing insulin-stimulated Akt phosphorylation in skeletal muscle, insulin (2 units/kg) was injected intraperitoneally in mice fasted for 5 h. Skeletal muscles (quadriceps) were harvested 15 min after insulin injection, snap frozen in liquid nitrogen, and stored at −80°C until homogenization.

Fat tissues were fixed in 4% buffered formalin and embedded in paraffin. Sections were cut and stained with hematoxylin and eosin. For each fat pad, at least 100 adipocytes were analyzed. Quadriceps femoris muscle was mounted in embedding medium (OCT embedding matrix, Cellpath Ltd., Newtown, U.K.), snap frozen in isopentane cooled to −160°C with liquid nitrogen, and subsequently stored at −80°C until use. Consecutive 12 µm sections were cut on a microtome at −25°C. Rat anti-mouse CD31 endothelial cell antibody (BioLegend, San Diego, CA) was used as a marker for muscle capillaries, and capillary-to-fiber ratio was calculated by dividing the number of CD31-positive cells by the number of muscle fibers. For all histochemical and immunohistochemical analyses, National Institutes of Health ImageJ software (National Institutes of Health, Bethesda, MD) was used.

Liver tissue (10–30 mg) was homogenized in PBS, and lipids were extracted in a chloroform-methanol (2:1) mixture. Total liver lipids were determined by a sulfophosphovanillin reaction as previously described (31). Liver triglycerides were measured from 50 mg of liver tissue using a variation of the Bligh and Dyer (32) method and quantified with an enzymatic assay (Roche Diagnostics, Rotkreuz, Switzerland).

Total RNA was extracted and reverse transcribed as described (33). The following primers were used: tumor necrosis factor-α Mm00443258_m1, interleukin (IL)-6 Mm00446190_m1, IL-1β Mm0043422/8_m1, cd11c Mm00498698_m1, adipose triglyceride lipase Mm00503040_m1, perilipin Mm00558672_m1, hypoxia-inducible factor (HIF)-1α Mm00468869_m1, CD31 Mm01242584_m1, and vascular endothelial growth factor (VEGF)-A Mm01281441_m1 (Applied Biosystems, Rotkreuz, Switzerland).

Cells or tissues were lysed and Western blots were performed as previously described (33). The following primary antibodies were used: anti-phospho-Akt (Ser473), anti-total Akt (Cell Signaling, Danvers, MA), anti-VEGF-A (Santa Cruz Biotechnology Inc., Dallas, TX), and anti-actin (Millipore, Zug, Switzerland).

Data are presented as mean ± SEM and were analyzed by unpaired Student t test. P values <0.05 were considered significant.

The surgical procedure (sham operation or UniNx) was performed in C57BL/6J mice at 7 weeks of age. Mice were fed either normal chow or an HFD for 20 weeks. Total body weight gain was similar in sham-operated and UniNx mice under both diets (Table 1). After 20 weeks of HFD, inguinal and mesenteric fat pad weights in the UniNx mice were comparable to sham-operated mice, whereas epididymal fat pad weight was significantly higher in UniNx mice (Table 1). No differences in fat pad weights were observed between chow-fed sham-operated and UniNx mice (Table 1). In addition, food intake, locomotion, or fuel utilization (respiratory quotient) were similar between HFD-fed sham-operated and UniNx mice (Supplementary Fig. 1).

Fasting blood glucose levels were similar between HFD-fed sham-operated and UniNx mice (Table 2). In addition, plasma insulin and FFA concentrations did not differ significantly between both groups (Table 2). Glucose and insulin tolerance were similar in UniNx mice compared with sham-operated mice under both standard chow as well as HFD 20 weeks after surgery (Fig. 1A and B). Hyperinsulinemic-euglycemic clamp studies revealed similar glucose infusion rate in HFD-fed UniNx compared with sham-operated mice (see Fig. 1C for steady state glucose infusion rates and Supplementary Fig. 2 for detailed time courses). Importantly, insulin-induced suppression of endogenous glucose production (mainly reflecting hepatic glucose production) was blunted in HFD-fed sham-operated mice but was clearly evident in UniNx mice (Fig. 1D), indicating improved/preserved hepatic insulin sensitivity in the latter. In contrast, glucose disposal rate during hyperinsulinemic-euglycemic clamp was significantly further deteriorated in HFD-fed UniNx mice, suggesting reduced skeletal muscle insulin sensitivity compared with sham-operated mice (Fig. 1E).

Hepatic insulin sensitivity is often associated with hepatic steatosis, though it is still debated whether insulin resistance is the cause of hepatic steatosis or whether the increase in triglycerides (or of lipid metabolites such as ceramides, diacylglycerol, and acyl-CoAs) causes the development of hepatic and/or systemic insulin resistance. As depicted in Fig. 2A, total liver lipid content was greatly reduced in HFD-fed UniNx compared with sham-operated mice. Similarly, liver triglyceride levels were reduced in HFD-fed UniNx mice (UniNx mice, 234.5 ± 41.0 µmol/g liver; sham-operated mice, 401.3 ± 77.5 µmol/g liver). In addition, histological examination of liver sections revealed reduced frequency of lipid vacuoles in HFD-fed UniNx mice (Fig. 2B).

The portal theory proposes that the direct exposure of the liver to increasing amounts of FFA and/or proinflammatory factors released from visceral fat into the portal vein may promote the development of hepatic insulin resistance and steatosis (34). Accordingly, mRNA expression of proinflammatory cytokines as well as adipose tissue histology was analyzed in order to characterize adipose tissue inflammation. As depicted in Fig. 2C, expression of CD11c (a marker for proinflammatory M1 polarized macrophages) and IL-6 was significantly decreased, and expression of TNF-α and IL-1β trendwise reduced in mesenteric adipose tissue of HFD-fed UniNx compared with sham-operated mice. Of note, mRNA levels of the HIF-1α were significantly lower in mesenteric adipose tissue of HFD-fed UniNx mice (Fig. 2C). In contrast, adipocyte size was not different between both groups of mice (Fig. 2D and E). In addition, insulin-stimulated Akt phosphorylation in adipose tissue was similar in both groups, suggesting similar insulin sensitivity in this tissue (Fig. 2F). Thus UniNx protects mice from the development of HFD-induced adipose tissue inflammation and consequently hepatic insulin resistance.

Glucose uptake into isolated soleus (slow twitch) and EDL (fast twitch) muscles was assessed next removing extramyocellular barriers to muscle glucose uptake. As depicted in Fig. 3A and B, basal and insulin-stimulated glucose uptakes were similar between HFD-fed sham-operated and UniNx mice, suggesting no direct impairment of insulin action at the myocyte level. Of note, 20 weeks of HFD led to an expected poor insulin response in skeletal muscle, which was particularly evident in soleus muscle. Similar to ex vivo glucose uptake, no difference between both groups of mice was found for in vivo insulin-stimulated Akt phosphorylation in total muscle homogenates (Fig. 3C). These data in isolated muscle contrasts with the data obtained from hyperinsulinemic-euglycemic clamp studies revealing diminished muscle insulin sensitivity in HFD-fed UniNx mice and may indicate that UniNx leads to impaired insulin delivery to the sarcolemma, e.g., through reduction in capillary density.

In order to assess capillary density, expression of the endothelial marker CD31 was determined. As depicted in Fig. 4A, mRNA levels of CD31 were significantly reduced in quadriceps muscle in HFD-fed UniNx compared with sham-operated mice, whereas no difference was observed in chow-fed mice (Supplementary Fig. 3). Moreover, capillary-to-fiber ratio was significantly (∼15%) reduced in skeletal muscle of HFD-fed UniNx mice (Figs. 4B and C). Thus UniNx reduces muscle capillary density in HFD-fed mice. Mechanistically, reduced muscle capillary density may be the result of reduced HIF-1α levels, which is a transcriptional activator of genes encoding VEGF and other important mediators of angiogenesis (35). Indeed, HIF-1α and VEGF-A mRNA levels were decreased in skeletal muscle of HFD-fed UniNx mice (Fig. 4D). Moreover, protein levels of VEGF-A were decreased by 40% in skeletal muscle of HFD-fed UniNx compared with sham-operated mice (Fig. 4E).

As reported in Table 2, angiotensin I levels were significantly increased in HFD-fed UniNx mice. Activation of the RAS was previously suggested to contribute to HFD-associated skeletal muscle insulin resistance (36–38). Thus we hypothesized that the observed increase in circulating angiotensin I concentration may deteriorate HFD-induced skeletal muscle insulin resistance via activation of angiotensin II type 1 receptors. To test such a hypothesis, we made use of the angiotensin II type 1 receptors blocker telmisartan. UniNx and sham-operated mice were fed an HFD for 20 weeks and received telmisartan in the drinking water at a dose of 3 mg/kg⋅day during the entire period. As depicted in Fig. 5A and B, telmisartan treatment improved glucose tolerance in HFD-fed sham-operated but not UniNx mice. Moreover, it improved glucose infusion rate (Fig. 5C) and hepatic insulin sensitivity (Fig. 5D) in sham-operated mice as analyzed by hyperinsulinemic-euglycemic clamp studies (see Supplementary Fig. 4 for detailed time courses). In accordance with elevated hepatic insulin sensitivity, telmisartan treatment reduced hepatic steatosis in sham-operated mice (total liver lipids in untreated HFD-fed sham-operated mice, 299.4 ± 59.6 mg/g liver tissue; total liver lipids in telmisartan-treated HFD-fed sham-operated mice, 158.0 ± 31.8 mg/g liver tissue) but did not affect it in UniNx mice (total liver lipids in untreated HFD-fed UniNx mice, 123.4 ± 17.7 mg/g liver tissue; total liver lipids in telmisartan-treated HFD-fed UniNx mice, 178.5 ± 22.8 mg/g liver tissue), resulting in similar total liver lipids in telmisartan-treated mice of both groups (Fig. 5E). Moreover, mRNA expression levels of proinflammatory cytokines in mesenteric adipose tissue were similar between telmisartan-treated HFD-fed sham-operated and UniNx mice (Fig. 5F). Of note, expression of HIF-1α was no longer different between telmisartan-treated sham-operated and UniNx mice (Fig. 5F). Such data suggest that telmisartan treatment improved adipose tissue inflammation in HFD-fed sham-operated mice since the observed difference in cytokine mRNA expression in untreated mice (Fig. 2A) was no longer present.

Previously, the angiotensin II receptor blocker losartan was found to reverse insulin resistance through the modulation of muscular circulation in rats with impaired glucose metabolism (20). We therefore postulated that telmisartan would improve muscle insulin sensitivity as well as capillary density in UniNx mice. However, as depicted in Fig. 6A, 20 weeks of telmisartan treatment did not affect IS-GDR in UniNx mice (IS-GDR in untreated HFD-fed UniNx mice, 10.6 ± 1.2 mg/kg⋅min; IS-GDR in telmisartan-treated HFD-fed UniNx mice, 4.2 ± 2.3 mg/kg⋅min). Accordingly, insulin-stimulated glucose uptake was significantly higher in telmisartan-treated HFD-fed sham-operated compared with UniNx mice (Fig. 6B). Moreover, CD31 expression remained reduced in the quadriceps muscle of telmisartan-treated HFD-fed UniNx compared with sham-operated mice (Fig. 6C). Thus telmisartan treatment positively impacts on total body and hepatic insulin sensitivity in HFD-fed sham-operated mice but does not affect impaired skeletal muscle insulin sensitivity in HFD-fed UniNx mice.

In the current study, we describe an unexpected dissociation of hepatic and muscle insulin resistance: although UniNx protected HFD-fed mice from the development of hepatic insulin resistance (and hepatic steatosis), it deteriorated skeletal muscle insulin resistance.

The observed association of reduced visceral adipose tissue inflammation and improved hepatic insulin sensitivity in HFD-fed UniNx mice is in accordance with the portal theory claiming that the exaggerated release of FFAs and/or proinflammatory cytokines from visceral fat are directly delivered to the liver via portal vein, promoting the development of hepatic insulin resistance and hepatic steatosis (34). Interestingly, expression of HIF-1α was reduced in adipose tissue of UniNx mice. Selective inhibition of HIF-1α was previously reported to reduce adipose tissue inflammation as well as hepatic steatosis in HFD-fed mice (39). Conversely, overexpression of HIF-1α induced adipose tissue fibrosis and inflammation as well as hepatic steatosis (40). Thus UniNx is associated with decreased HIF-1α expression in adipose tissue and thereby may contribute to reduced adipose tissue inflammation and, consequently, preserved hepatic insulin sensitivity. Such a notion is further supported by the fact that treatment with telmisartan abolished differences in mRNA expression of HIF-1α as well as proinflammatory cytokines and improved hepatic insulin sensitivity in sham-operated mice.

The kidney contributes a significant part (∼15–20%) to endogenous glucose production in both humans and rats (41,42). Thus basal endogenous glucose production may be reduced in UniNx mice. However, similar basal endogenous glucose production was observed in HFD-fed UniNx and sham-operated mice (UniNx, 25.8 mg/kg⋅min; sham-operated, 22.2 mg/kg⋅min; P = 0.27). Moreover, insulin-mediated suppression of endogenous glucose production was increased in HFD-fed UniNx mice. Since renal glucose production is inhibited by insulin, we cannot exclude that such improvement may be due to improved renal insulin sensitivity in UniNx mice. However, the fact that adipose tissue inflammation as well as liver lipid accumulation was decreased in HFD-fed UniNx mice and the notion that endogenous glucose production mainly reflects hepatic glucose production render improved hepatic insulin sensitivity a more likely explanation herein.

Decreased mRNA expression and histological staining of CD31 point toward reduced capillary density in skeletal muscle of HFD-fed UniNx mice. Capillary rarefaction was previously associated with insulin resistance both in humans and rodents (19–21). In accordance with our findings presented herein, Flisiński et al. (43) reported reduced numbers of capillaries as well as a decreased capillary-to-fiber ratio in gastrocnemius and longissimus muscle in UniNx male Wistar rats. Such changes were already present at an early stage and independent of blood pressure. Of note, treatment with the angiotensin II receptor blocker telmisartan had no impact on capillary density in UniNx mice, suggesting that UniNx-associated capillary rarefaction is not dependent on activation of the RAS, whereas it may modulate muscular circulation and thereby insulin sensitivity as was previously reported for the angiotensin II receptor blocker losartan in rats (20). Herein, reduced capillary density was associated with decreased HIF-1α mRNA expression in skeletal muscle of UniNx mice. HIF-1α is the master regulator of transcriptional responses to hypoxia and a transcriptional activator of genes encoding VEGF and other important mediators of angiogenesis (35). Indeed, VEGF-A mRNA and protein levels were reduced in HFD-fed UniNx compared with sham-operated mice, suggesting that reduced HIF-1α may be responsible for reduced capillary density in the former. Importantly, UniNx and subtotal nephrectomy in Wistar rats were associated with reduced expression of HIF-1α, VEGF-A, and VEGF receptor in gastrocnemius muscle (44). In addition, subtotal nephrectomy reduced skeletal muscle angiogenesis in rats (45). Unfortunately, HIF-1α expression was not determined in the latter publication. Moreover, muscle-specific deletion of VEGF resulted in capillary rarefaction and diminished insulin-induced muscle glucose uptake in vivo independent of defects in insulin action at the myocyte (22). Thus reduced expression of HIF-1α may be causally linked to reduced capillary density observed in skeletal muscle of UniNx mice.

In conclusion, UniNx unexpectedly protects mice from HFD-induced adipose tissue inflammation and hepatic insulin resistance potentially via a reduction in HIF-1α expression. In contrast, UniNx leads to capillary rarefaction in skeletal muscle and, thus, deteriorates HFD-induced skeletal muscle glucose disposal in vivo.

Acknowledgments. The authors acknowledge Giatgen Spinas, University Hospital Zurich, for continuous support and Dr. Denis Arsenijevic as well as Jean-Pierre Montani, University of Fribourg, for expert advice regarding the UniNx procedure.

Funding. This work was supported by grants from the Swiss National Center for Competence in Research (NCCR-Kidney.ch to G.A.K.-U. and D.K.), the Swiss National Science Foundationhttp://dx.doi.org/10.13039/501100001711 (310030-141238 to D.K.), the Foundation for Research at the Medical Faculty, University of Zurich (to F.I.), and the International Fellowship Program (grant 246539) on Integrative Kidney Physiology and Pathophysiology (to G.A.K.-U.). M.S.F.W. was supported by Wolfermann-Nägeli-Stiftung. Telmisartan was kindly provided by Boehringer Ingelheim International GmbH (MTA 1037).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. S.H.C., F.I., S.W., Z.Z., M.S.F.W., and Z.G. performed the experimental work. E.J.S., G.A.K.-U., and H.A.-H. gave conceptual advice. D.K. conceived the study and wrote the manuscript. All authors contributed to discussion and reviewed/edited manuscript. D.K. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.